PRIYANKA VERGADIA: Welcome
to Deconstructing Chatbots. I'm Priyanka Vergadia. And in this episode,
we will learn how to integrate a Dialogflow
agent with a back end system like a database,
calendar, or a CRM. [MUSIC PLAYING] Any chatbot you build will
need to connect to a back end to transfer information to
and from the business logic. Within Dialogflow, we are able
to make that connection using fulfillment. Fulfillment is a piece of code
which is deployed as a web hook and lets your Dialogflow
agent call the business logic on a per intent basis. During a conversation,
this allows you to use the information
extracted from Dialogflow's natural language processing
to generate a dynamic response or to trigger action
on your back end. Let's take our appointment
scheduler chatbot from previous episodes and
implement a fulfillment for scheduling appointments. Well, currently, our agent
identifies the user's intent of scheduling an
appointment and provides a static response indicating
that the appointment is set up. Now, in a real
world scenario, we need to set up that
appointment on a calendar and also check for any
scheduling conflicts, which requires us to integrate
with the calendar. So let's jump into
our Dialogflow console and see how it is all set up. Click on the gear icon near your
agent name to open Settings. Look at the Google
Cloud Project ID and click on it to open
the project in GCP console. Navigate to the
hamburger menu icon and locate APIs and Services. Click on Enable
APIs and Services and look for Google
Calendar API. Once you find it, enable the
API on your cloud project. Now under the same APIs and
Services within Credentials, click on Create
credentials dropdown to select Service account key. Under Create
service account key, select the New service
account from the dropdown and enter an account name. Make sure JSON is the key
type and then click Create. In the popup, select
Create without role. A JSON file will be
downloaded to your computer. And you will need this later,
so take note of its location. Before we close it out
though, copy the client email. Now open Google Calendar. And on the left, click the plus
sign and select New Calendar. Enter AppointmentCalendar
for the name of the calendar and select Create Calendar. Next, go to the
AppointmentCalendar that will appear on the left. Click Add People
and paste the email copied from that JSON file. Then select Make changes
to events in the Permission dropdown and click Send. At this point, we are ready to
go back to our Dialogflow agent and set the fulfillment. Click on Fulfillment and
enable the inline editor. Check out the link
in the description below to get the required code
for package.json and index.js. Open package.json file and
copy and paste the dependencies into fulfillment. Do the same for
index.js as well. Grab the Calendar ID
from the Google Calendar. We just create it and
replace that in index.js. Next, copy the contents of
the JSON file downloaded in the Service
Account Setup section and paste it into the empty
object, ServiceAccount. Check out the rest of the code. Make appointment function
contains the responses to the user. And create calendar
event function takes care of grabbing the
entities like date, time, and appointment type, and
creates the calendar invite. It also looks for the
calendar conflicts and throws exceptions. Once you're done, click Deploy
at the bottom of the page. Now, we have one more thing to
do before we can start testing. Click on Intents and go to the
Schedule Appointment intent. Scroll down to Fulfillment
and enable webhook and save. Now we're ready to test. In the simulator on the right,
query your Dialogflow agent with something like, set
an appointment at 4:00 PM tomorrow. And then respond to
the follow up queries. After getting the
required information, you will see that the
appointment is added to the appointment calendar. You can verify that if you
ask for another appointment at the same time slot, your
agent will respond back with an appropriate message
since that slot is not free. All right. So that was a lot. Let's summarize what
we learned today. We enhanced our appointment
scheduler chatbot by creating a fulfillment
through inline editor, enabled Google Calendar
API for the GCP project and used those APIs
credentials to connect our fulfillment Cloud
Function with Google Calendar. And finally, we
were able to test our agent by
creating appointment request on the calendar. Don't miss the next episode
of Deconstructing Chatbots to learn more about
conversational AI. Please like and subscribe to our
channel for more such content. [MUSIC PLAYING]
